Re: Chi-Square Determinations

[tdc@ehssun.lbl.gov (Ted M. de Castro)]

>>Some training material I was reviewing several years ago suggested that 
>>one  should not subtract background from source measurements made for 
>>chi-square  determinations.  It was stated that this would add a bias to 
>>the results. 
  
>>I've not been able to locate a reference for this.
     
I would also like to see such a reference since this is a particularly hot 
issue for us right now too.

For us it is compounded by the fact that terrestrial background can change 
substantially from one place to another.

Another concern is statistical variance where it is important to do the 
statistics on the raw data and not derived quantities - or have a good 
understanding the units and calculations so as to make the proper and valid 
assessment.


Ted de Castro
tdc@ehssun.lbl.gov
University of California Lawrence Berkeley Laboratory
Bldg B75B Rm 101
Berkeley, CA 94720
(510) 486-5256
(510) 486-7304 - FAX